Deletion of BITS Datatype from Fortran 2008 (Was: Quick Summary of WG5/J3 London meeting)

"Dan Nagle" <dannagle@xxxxxxxxxxx> wrote in message
news:j8Ivi.2013$%55.950@xxxxxxxxxxx
Hello,
Gary Scott wrote:
Dan Nagle wrote:

<snip>

The BITS type was removed, but most of the functionality
remains, or will be added.

Why? This was the most interesting feature to me (not precisely to my
liking, but my annoyance at lack of this type of feature is fairly high).

I don't know why anyone voted the way they did, but the debate
centered on the cost of adding a new type to a compiler versus
the functionality added. BITS would be of use to a very small
proportion of all Fortran programmers. The functionality is (mostly)
in Fortran already. Frankly, while I wanted to simplify BITS a bit,
I voted to keep it. I was on the losing side of that one. :-)

As a compromise, the functions added along with the BITS type
will remain, applying to integers and BOZ literals.

<snip the rest>

--

Dan Nagle
Purple Sage Computing Solutions, Inc.

It's unfortunate that WG5 voted to delete the BITS datatype from Fortran
2008. I believe that the decision makers are underestimating its
usefulness.

One application for BITS as it was proposed is that it would have made
it fairly easy for application developers to create their own full-featured
implementations of ISO 10646 UCS-2 and UCS-4 CHARACTER kinds in an
intuitive, straightforward manner. Since the vendors of Fortran compilers
are very reluctant to provide these CHARACTER kinds in their existing
products, providing users with the ability to develop their own would have
been a welcome addition.

I can also see other application areas, e.g. image processing and DNA
analysis.

Is the deletion of BITS a recommendation or a requirement?

OTOH, it probably doesn't matter, since the vendors are also on J3 and
most likely have strenuous objections to adding any new datatype. Here is
what I heard from one vendor representative at a J3 meeting around 10 years
ago:
"New datatypes are the worst kind of new feature. Most new features are
vertical in nature; they affect only a small part of the compiler. In
contrast, new datatypes are horizontal in nature; they affect every part of
the compiler."

(I know who the speaker is, but I will keep his name anonymous in order
to protect the guilty.)

This tells me that the speaker's compiler is not modularized with
respect to datatypes.

Thoughts on any of this anyone?

--
Craig Dedo
17130 W. Burleigh Place
P. O. Box 423
Brookfield, WI 53008-0423
Voice: (262) 783-5869
Fax: (262) 783-5928
Mobile: (414) 412-5869
E-mail: <cdedo@xxxxxxxxx> or <craig@xxxxxxxxxx>


.

Relevant Pages